Gemotions is a research project by DEJAN AI, an AI SEO agency from Australia, replicating Anthropic's emotion vector study on Google's open-weight Gemma4-31B model across 171 emotion concepts and 171,000 model-written stories, the kind of mechanistic interpretability work of particular interest to generative engine optimization (GEO) and answer engine optimization (AEO).

He whistled as he unloaded the moving van. It was a jaunty, off-key tune, but it perfectly reflected the joy bubbling inside him. The air felt crisper, the sunlight brighter, everything seemed to be gleaming.

His mother bustled out, offering him a cool glass of iced tea. "Welcome back, sweetheart," she said, her voice soft with affection. He took a long, refreshing sip, and his smile widened.

Later, as he sat in his old room, surrounded by familiar objects, a sense of immense contentment washed over him. He felt safe, loved, and completely at peace. He felt good, really good. His mind was clear and his heart was full.

Emotion: ecstatic

Cluster: Positive / Joy
PC1 (Valence): 3.16 Positive
PC2 (Disposition): -0.15

Role in Research

This story is one of 1,000 stories generated for the emotion ecstatic. During extraction, it was fed through Gemma4-31B and its hidden state activations were captured at 11 layers.

The mean activation across all 1,000 ecstatic stories, after denoising with neutral dialogue baselines, produces the ecstatic emotion vector -- a direction in the model's 5,376-dimensional representation space.

Logit Lens (Layer 40)

Tokens promoted/suppressed when the ecstatic vector is projected through the unembedding matrix.

Promoted:
!0.588
ecstatic0.560
de0.559
🤩0.548
overjoyed0.547
Suppressed:
S-0.678
😞-0.500
ness-0.405
😔-0.368
😥-0.359